US20120084522A1

US20120084522A1 - Virtualization control apparatus and storage system

Info

Publication number: US20120084522A1
Application number: US13/239,812
Authority: US
Inventors: Hiroshi Shiomi; Koutarou Sasage; Akira Satou; Ryosuke Suzuki; Yasuhito Kikuchi; Kenichi Fujita
Original assignee: Fujitsu Ltd
Current assignee: Fujitsu Ltd
Priority date: 2010-10-04
Filing date: 2011-09-22
Publication date: 2012-04-05
Also published as: JP5636853B2; US8683155B2; JP2012079183A

Abstract

A virtualization control apparatus includes a selection unit that, when receiving a copy request, conducting an access test corresponding to the copy request on each of the virtualization switch units and selects one of the virtualization switch units of the highest performance among the plurality of the virtual switch units, and a sending unit that sends the copy request to the selected virtualization switch unit.

Description

CROSS-REFERENCE TO RELATED APPLICATION

This application is based upon and claims the benefit of priority of the prior Japanese Patent Application No. 2010-225257, filed on Oct. 4, 2010, the entire contents of which are incorporated herein by reference.

FIELD

The embodiment discussed herein is related to a virtualization control apparatus and a storage system.

BACKGROUND

Storage virtualization switches (hereafter referred to as virtualization switch devices) are devices for integrally controlling physical volumes of multiple disk units and showing the physical volumes to the host computer as logical volumes. Such a virtualization switch device is commonly connected to a disk device including multiple disk units and to a server that makes a request to the virtualization switch devices, via lines such as fiber channels.
Among disk devices is one including a control module (hereafter referred to as a CM) that controls multiple disks under its command.
Connected to such a CM via disk adaptors (DA) or the like are multiple disk units. The virtualization switch device accesses the disk units via the CM.
Such storage systems include one that makes the virtualization switch device redundant by including multiple virtualization switch devices and enabling a controller, the server, to access each virtualization switch device. This configuration allows reducing the load imposed on each virtualization switch device in the storage system, as well as continuously operating the entire system even when any of the virtualization switch devices goes down.
Accordingly, in the related-art controller, it is difficult to distribute the load in accordance with the load situation of the actual CMs or disk units, although it is possible to distribute the load corresponding to the logical volumes. In particular, in a copy process, reading of data and writing of the data are performed on multiple disk units. Thus, depending on which of the virtualization switch devices the server instructs to perform the copy process, performance often significantly varies.
Japanese Laid-open Patent Publication (Translation of PCT Application) No. 2005-505035

SUMMARY

According to an aspect of the embodiment, a virtualization control apparatus of a storage system, the storage system including a plurality of storage controllers controlling a plurality of storage units connected thereto and a plurality of virtualization switch units being connected to one of the storage controllers, each of the virtualization switch units converting a copy source logical volume address and a copy destination logical volume address contained in a received copy request into addresses of the storage units and making a request to the storage controllers on the basis of the converted addresses. The virtualization control apparatus includes a selection unit that, when receiving a copy request, conducting an access test corresponding to the copy request on each of the virtualization switch units and selects one of the virtualization switch units of the highest performance among the plurality of the virtual switch units, and a sending unit that sends the copy request to the selected virtualization switch unit.
The object and advantages of the invention will be realized and attained by means of the elements and combinations particularly pointed out in the claims.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are not restrictive of the invention, as claimed.

BRIEF DESCRIPTION OF DRAWINGS

FIG. 1 is a block diagram of a storage system according to an embodiment.

FIG. 2 is a diagram illustrating a logical volume management table according to this embodiment.

FIG. 3 is a diagram illustrating an address conversion table according to this embodiment.

FIGS. 4A to 4D are diagrams illustrating a request from a device according to this embodiment.

FIG. 5 is a diagram illustrating a RAID management table according to this embodiment.

FIG. 6A and 6B are diagrams illustrating a write request according to this embodiment.

FIG. 7 is a diagram illustrating example processes performed by the storage system according to this embodiment.

FIG. 8 is a diagram illustrating example processes performed by the storage system according to this embodiment.

FIG. 9 is a diagram illustrating example processes performed by the storage system according to this embodiment.

FIG. 10 is a diagram illustrating example processes performed by the storage system according to this embodiment.

FIG. 11 is a diagram illustrating example processes performed by the storage system according to this embodiment.

FIG. 12 is a diagram illustrating example processes performed by the storage system according to this embodiment.

FIG. 13 is a flowchart illustrating processes when receiving a copy request according to this embodiment.

FIG. 14 is a flowchart illustrating a process of selecting a virtualization switch device in charge of copy according to this embodiment.

FIG. 15 is a diagram illustrating a virtualization switch device-in-charge management table according to this embodiment.

FIG. 16 is a flowchart illustrating a monitoring process according to this embodiment.

FIG. 17 is a diagram illustrating a virtualization switch device-in-charge management table according to this embodiment.

FIG. 18 is a diagram illustrating a virtualization switch device-in-charge management table according to this embodiment.

DESCRIPTION OF EMBODIMENT

In the above-mentioned related-art configuration, the virtualization switch devices convert the physical volumes of the disk units into logical volumes. Accordingly, the controller may recognize only information corresponding to the logical volumes. For this reason, when the server instructs a virtualization switch device to access a certain logical volume, it may not know which of the disk units the virtualization switch device is accessing via which CM.
FIG. 1 is a diagram illustrating the configuration of a storage system according to an embodiment.
The storage system according to this embodiment includes a server 1 as a controller, virtualization switch devices 2 a and 2 b, and a disk device 3.
The server 1 accesses data stored in the disk device 3 in accordance with a request a terminal 4 sends via a network 5. The server 1 includes HBAs 101 a and 101 b, which are interfaces used to communicate with the virtualization switch devices 2 a and 2 b. The server 1 also includes a communication adaptor 104, which is connected to the network 5 and used to communicate with the terminal 4.
The server 1 also includes a processor 102, which controls components within the server. The processor 102 controls the HBAs 101 a and 101 b to make to both the virtualization switch devices 2 a and 2 b a write request, read request, or copy request to be sent to virtual logical volumes.
The server 1 also includes a storage unit 106. Unless otherwise specified, it is assumed in this embodiment that the processor 102 performs processes to be discussed later by executing programs stored in the storage unit 106.
The virtualization switch devices 2 a and 2 b integrally controls the disk device 3 controlled by multiple CMs, 30 a and 30 b and perform a process for illustrating physical volumes to the host computer, the server 1, as virtual volumes. The virtualization switch devices 2 a and 2 b includes the same components so as to obtain a redundant configuration.
The virtualization switch device 2 a includes ports 201 a 1 to 201 a 5, a switch 202 a, a processor 203 a, and a storage unit 204 a. The virtualization switch device 2 b includes ports 201 b 1 to 201 b 5, a switch 202 b, a processor 203 b, and a storage unit 204 b.
Connected to the ports 201 a 1 to 201 a 5 and 201 b 1 to 201 b 5 are communication lines from other devices. The virtualization switch devices 2 a and 2 b receive or send data from or to the other devices via the communication lines.
In this embodiment, the HBAs 101 a and 101 b of the server 1 are connected to the ports 201 a 1 and 201 b 1, respectively. The port 201 a 4 of the virtualization switch device 2 a is connected to a CA 301 a 1 of the CM (controller) 30 a of the disk device 3. Likewise, the port 201 b 4 of the virtualization switch device 2 b is connected to a CA 301 b 1 of the CM (controller) 30 b of the disk device 3.
The switches 202 a and 202 b are connected to the ports 201 a 1 to 201 a 5 and 201 b 1 to 201 b 5, respectively. The switch 202 a transfers data received by one of the ports 201 a 1 to 201 a 5 to the other ports under the control of the processor 203 a. Likewise, the switch 202 b transfers data received by one of the ports 201 b 1 to 201 b 5 to the other ports under the control of the processor 203 b.
Specifically, the processor 203 a converts a disk address attached to data received from one of the ports 201 a 1 to 201 a 5 on the basis of an address conversion table 210 a stored in the storage unit 204 a and then controls the switch 202 a to transfer the converted data to a port (one of the 201 a 1 to 201 a 5) to which the destination device is connected. Likewise, the processor 203 b converts a disk address attached to data received from one of the ports 201 b 1 to 201 b 5 on the basis of an address conversion table 210 b stored in the storage unit 204 b and controls the switch 202 b to transfer the converted data to a port (one of 201 b 1 to 201 b 5) to which the destination device is connected.
The storage units 204 a and 204 b are storing the logical volume management tables 205 a and logical volume management table 205 b, respectively, which show the correspondences between the logical volumes and the virtualization switch devices 2 a and 2 b.
As illustrated in FIG. 2, logical volume numbers LV-01 and LV-02 and the numbers of the virtualization switch devices are stored as associated with each other in the logical volume management tables 205 a and 205 b. For the sake of convenience, it is assumed in this embodiment that the numbers of the virtualization switch devices are “2 a”“2 b,” which are the same as their symbols.
The storage units 204 a and 204 b also contain virtualization switch device-in-charge management tables 206 a and 206 b, respectively, storing information as to which of the virtualization switch devices 2 a and 2 b is in charge of each of multiple copy processes. As will be described later, in a copy process, a selection process with reference to the virtualization switch device-in-charge management tables 206 a and 206 b takes precedence over a selection process with reference to the logical volume management tables 205 a and 205 b. The virtualization switch device-in-charge management tables 206 a and 206 b will be described in detail later.
The storage units 204 a and 204 b also contain address conversion tables 210 a and 210 b, respectively, as illustrated in FIG. 3.
As illustrated in FIG. 3, logic addresses in the logical volumes LV-01 and LV-02 and the addresses of disk units 305-1, 305-3, 305-5, and 305-7 among disk units 305-1 to 305-8 stored in the disk device 3 are stored as associated with each other in the address conversion tables 210 a and 210 b.
The processors 203 a and 203 b convert addresses with reference to the address conversion tables 210 a and 210 b. For example, when the port 201 a 1 receives from the server 1 a request 1001 for reading 020010h of the logical volume LV-01 as illustrated in FIG. 4A, the processor 203 a extracts the address of a disk unit corresponding to the logical volume with reference to the address conversion table 210 a. That is, the processor 203 a extracts an address 00020h of the disk unit 305-3. As illustrated in FIG. 4B, the processor 203 a generates a read request 2001 by converting the destination of the read request 1001 into the extracted disk unit address. The processor 203 a then sends the request 2001 to the disk device 3 via the port 201 a 4.
Subsequently, when the switch 202 a receives a response 3001 (FIG. 4C) corresponding to the read request 2001 from the disk device 3, the processor 203 a converts the address with reference to the address conversion table 210 a.
Specifically, the processor 203 a generates a response 2002 (FIG. 4D) by converting the source address of the response 3001 into an address of the logical volume LV-01 with reference to the address conversion table 210 a.
The processor 203 a then sends the response 2002 to the destination, the disk device 1, via the port 201 a 1.
Note that, in this embodiment, the address conversion table 210 a contained in the storage units 204 a and the address conversion table 210 b contained in the storage unit 204 b are storing the same values. Further, it is assumed that the processors 203 a and 203 b of the virtualization switch devices 2 a and 2 b communicate with each other via the ports 201 a 6 and 201 b 6 so that the same values are always held in the address conversion tables 210 a and 210 b.
The storage units 204 a and 204 b also contain RAID management tables 211 a and 211 b, respectively.
As illustrated in FIG. 5, the RAID management tables 211 a and 211 b are storing the correspondences between the disk units 305-1, 305-3, 305-5, and 305-7 and the disk units 305-2, 305-4, 305-6, and 305-8.
When the 203 a and 203 b receive a write request from the server 1 as described above, they perform the above-mentioned conversion process, as well as generate a write request to be sent to a disk unit serving as a mirror with reference to the RAID management tables 211 a and 211 b.
For example, when the processor 203 a generates a write request 2003 to be sent to the disk unit 305-3 illustrated in FIG. 6A by converting the address of the request from the server 1, it refers to the RAID management table 211 a.
In the RAID management table 211 a, a mirror corresponding to the disk unit 305-3 is the disk unit 305-4. Accordingly, as illustrated in FIG. 6B, the processor 203 a generates a request 2004 by converting the destination of the request 2003 into the disk unit 305-4 and sends the request 2004 to the disk device 3.
Back in FIG. 1, the disk device 3 is connected to the virtualization switch devices 2 a and 2 b.
The disk device 3 includes the multiple disk units, 305-1 to 305-8, and the CMs (controller modules) 30 a and 30. The CM 30 a includes a CA (channel adaptor) 301 a 1 connected to the port 201 a 4 of the virtualization switch device 2 a. Likewise, the CM 30 b includes a CA 301 b 1 connected to the port 201 b 4 of the virtualization switch device 2 b. The CM 30 a includes DAs (disk adaptors) 303 a 1 to 303 a 4 connected to the disk units 305-1 to 305-4. Likewise, the CM 30 b includes DAs 303 b 1 to 303 b 4 connected to the disk units 305-5 to 305-8.
The CMs 30 a and 30 b are connected to each other via the CAs 301 a 2 and 301 b 2 so that they may communicate with each other.
The switch 302 a controls the communication between the CMs 301 a 1 and 301 a 2 and the DAs 303 a 1 to 303 a 4 under the control of the processor 304 a. Likewise, the switch 302 b controls the communication between the CMs 301 b 1 and 301 b 2 and the DAs 303 b 1 to 303 b 4 under the control of the processor 304 b.
When the processor 304 a receives a request from the virtualization switch device 2 a via the CA 301 a 1, it controls the switch 302 a to write data to one of the disk units 305-1 to 305-4. Likewise, when the processor 304 b receives a request from the virtualization switch device 2 b via the CA 301 b 1, it controls the switch 302 b to write data to one of the disk units 305-5 to 305-8.
When the CM 30 a or CM30 b receives a request about one of the disk units 305-1 to 305-8 not under its charge, the processor 304 a or processor 304 b controls the switch 302 a or switch 302 b to transfer the request to the other CM via the CA 301 a 2 or CA 301 b 2.
For example, assume that the CA 301 a 1 of the CA 30 a has received from the virtualization switch device 2 a a read request to be sent to the disk unit 305-7. Since none of the DAs 303 a 1 to 303 a 4 is not connected to the disk unit 305-7, the processor 304 a controls the switch 302 a to transfer the request to the CM 30 b via the CA 301 a 2. When the CA 301 b 2 of the CM 30 b receives the transferred request, the processor 304 b controls the switch 302 b so that the request is relayed to the DA 303 b 3 connected to the disk unit 305-7.
A copy process performed on the logical volume LV01 in the storage server system configured as described above will be described.
As described above, the virtualization switch devices 2 a and 2 b may perform a process using logical volumes by performing conversion between a logical volume number and an address of the disk device. However, the logical volumes and the disk units of the disk device do not correspond to each other one-to-one as in this embodiment. Further, as illustrated in the logical volume number LV-02 of the address conversion tables 210 a and 210 b of FIG. 3, a single logical volume may correspond to the disk unit 305-1 controlled by the CM 30 a and the disk units 305-3 and 305-5 controlled by the CM 30 b.
The server 1 sends a request to both the virtualization switch devices 2 a and 2 b via the HBAs 101 a and 101 b.
The processor 203 a of the virtualization switch device 2 a determines whether the request is a request the processor is in charge of, with reference to the logical volume management table 205 a stored in the storage unit 204 a. Likewise, the processor 203 b of the virtualization switch device 2 b determines whether the request is a request the processor is in charge of, with reference to the logical volume management table 205 b stored in the storage unit 204 b. For example, the logical volume management tables 205 a and 205 b of FIG. 2 illustrate that LV-01 and the virtualization switch device 2 a correspond to each other. Accordingly, if the request is a request sent to LV-01, the processor 203 a of the virtualization switch device 2 a determines that the processor is in charge of the request. Note that since the virtualization switch device 2 a communicates with the disk device 3 via the CM 30 a, the request may be transferred from the CM 30 a to the CM 30 b depending on its destination address in the logical volume LV-01.
In the meantime, the server 1 only specifies the logical volume LV-01 or logical volume LV-02 as the destination of a request and does not control actual access to the disk units 305-1 to 305-8. Accordingly, it does not know which of the CMs 30 a and 30 b is being accessed.
Cases where the terminal 4 makes a request to a logical volume in this state will be described.
First, a case where a request received by the CM 30 a or CM 30 b is to write data to a disk unit under its charge will be described.
This case will be described specifically with reference to FIG. 7, assuming that the terminal 4 has sent a write request to “000100h” of “LV-01” to the server 1.
The processor 102 of the server 1 receives this request via the communication adaptor 104 and sends the write request to both the virtualization switch devices 2 a and 2 b via the HBAs 101 a and 101 b ( arrows 701 a and 701 b).
The ports 201 a 1 and 201 b 1 of the virtualization switch devices 2 a and 2 b receive the write requests and transfer them to the switches 202 a and 202 b ( arrows 702 a and 702 b). When the processors 203 a and 203 b receive the requests via the switches 202 a and 202 b, they determine which of the virtualization switch devices 2 a and 2 b is in charge of this request, with reference to the logical volume management tables 205 a and 205 b. In this embodiment, the virtualization switch device 2 a is in charge of the request, so the processor 203 b does not perform the subsequent processes.
The processor 203 a, which is in charge of the request, converts the destination address in accordance with the above-mentioned procedure with reference to the address conversion table 210 a.
In this embodiment, the address conversion table 210 a illustrated in FIG. 3 is stored. Accordingly, the processor 203 a generates a write request by converting the destination of the write request into “000100h” of the disk unit 305-1 with reference to this table. The processor 203 a then controls the switch 202 a to send the generated write request to the disk device 3 ( arrows 703 a and 704 a).
Here, the processor 203 a also refers to the RAID management table 211 to generate a similar write request to be sent to an address 000100h of the disk unit 305-2 serving as a mirror. The processor 203 a then controls the switch 202 a to send this write request to the disk device 3 ( arrows 703 b and 704 b).
The CA 301 a 1 of the CM30 a of the disk device 3 receives these requests and transfers them to the switch 302 a (arrows 705 a and 705 b). In accordance with these requests, the processor 304 a controls the switch 302 a to send the write requests to the destination disk units (arrow 706 a→707 a, arrow 706 b-707 b).
As seen, when the CM30 a or CM30 b receives a request to be sent to the disk units 305-1 to 305-4 or disk units 305-5 to 305-8 under its command, no load is imposed on the virtualization switch device or CM that has not received the request (in the above case, the virtualization switch device 2 b and CM 30 b).
Next, a case where a request received by the CM 30 a or CM 30 b is to write data to a disk unit not under its command will be described.
This case will be described specifically with reference to FIG. 8, assuming that the terminal 4 has sent a write request to “000100h” of “LV-02” to the server 1.
In this case, the server 1 sends the write request to the virtualization switch devices 2 a and 2 b ( arrows 801 a and 801 b), and the processor 203 b of the virtualization switch device 2 b generates a write request by converting the destination of the write request into “040100h” of “305-3” and sends the generated write request to the CM 30 b ( arrows 803 a and 804 a). The destination of a mirror write request is “040100h” of “305-4.” This write request is also sent to the CM 30 b (arrows 803 b and 804 b).
When the CA 301 b 1 receives these write requests, the processor 304 b refers to the destinations of the write requests. The processor 304 b then recognizes that the destinations are not any of the disk units 305-5 to 305-8 under its command and controls the switch 302 b to sends the write requests to the CM 30 a via the CA 301 b 2 ( arrows 805 a and 805 b, arrows 806 a and 806 b).
The CA 301 a 1 of the CM30 a receives these write requests and transfers them to the switch 302 a ( arrows 807 a and 807 b). The processor 304 a then controls the switch 302 a to transfer the write requests to the destination disk units ( arrows 808 a and 809 a, arrows 808 b and 809 b).
As illustrated in this example, when the CM 30 a or CM30 b receives requests to be sent to the disk units 305-5 to 305-8 or disk units 305-1 to 305-4 not under its command, a load is imposed on both the CMs 30 a and 30 b.
Next, a copy process performed between disk units under the command of the same CM will be described.
Such a copy process will be described with reference to FIG. 9 using a case where the terminal 4 has sent to the server 1 a request for copying data in “000000h” of “LV-01” to “040000h” of LV-01. Note that the server 1 has received the copy request and sent it to the virtualization switch devices 2 a and 2 b by controlling the HBAs 101 a and 101 b (arrows 901 a and 901 b).
A copy process includes a process of reading data in the copy source and a process of writing the read data to the copy destination. In this embodiment, a virtualization switch device corresponding to the copy source logical volume is in charge of the read and write processes.
The processors 203 a and 203 b of the virtualization switch devices 2 a and 2 b receive the copy request via the ports 201 a 1 and 201 b 1 and the switches 202 a and 202 b thereof (arrows 902 a and 902 b) and determine which of the virtualization switch devices 2 a and 2 b is in charge of the request, with reference to the virtualization switch device-in-charge management tables 206 a and 206 b. This determination will be described later. Here, assume that they have determined that the virtualization switch device 2 a is in charge. In this case, the processor 203 b, which is not in charge, does not perform the subsequent processes. On the other hand, the processor 203 a converts the address of the copy source logical volume contained in the copy command into an address of a corresponding disk unit with reference to the address conversion table 210 a. In this embodiment, the logical volume address is converted into “000000h” of the “disk unit 305-1.”
The processor 203 a then generates a read request corresponding to the converted address and controls the switch 202 a to send the read command to the CM 30 a (903, 904).
The processor 304 a receives this read command via the CA 301 a 1 and the switch 302 a and controls the switch 302 a to send the read command to the corresponding disk unit 305-1.
The disk unit 305-1 reads the data, and the read data is transferred to the switch 202 a of the virtualization switch device 2 a in accordance with a procedure reverse to the arrows 903 to 906 (907 to 912). The processor 203 a previously converts the address of the copy destination contained in the copy request from the server 1 using the address conversion table 210 a and generates a request for writing the data received by the switch 202 a to the converted address. The processor 203 a then controls the switch 202 a to send the generated write request to the CM 30 a. This process is similar to the process illustrated by the arrows 704 a and 704 b to the arrows 707 a and 707 b of FIG. 7 and will not be described (913 a and 913 b to 917 a and 917 b).
As seen, it is understood that a high load is imposed on the communication path from the switch 202 a of the virtualization switch device 2 a in charge of copy to the switch 302 a within the CM 30 a having the disk units to be copied under its command, as well as on the devices involved.
As described in FIG. 8, when the virtualization switch device 2 a or virtualization switch device 2 b makes a request to the CM 30 b or CM 30 a not under its command, it sends the request to the CM 30 b or CM 30 a via the CM 30 a or CM 30 b under its command.
The same goes for copy. That is, as described in FIG. 10, when the virtualization switch device 2 a or virtualization switch device 2 b makes a request to the CM 30 b or CM 30 a not under its command, it communicates with the CM 30 b or CM 30 a via the CM 30 a or CM 30 b under its command.
This increases the communication load between the CMs 30 a and 30 b.
FIG. 11 illustrates a case where a copy process that a virtualization switch device performs on a CM under its command, illustrated in FIG. 9, and a copy process that another virtualization switch device performs on a CM not under its command, illustrated in FIG. 10, are performed simultaneously. As illustrated in FIG. 11, a higher load is imposed on the switch 302 a of the particular CM (in FIG. 11, CM 30 a).
FIG. 12 illustrates a case where the virtualization switch devices 2 a and 2 b simultaneously perform copy processes on the CMs 30 b and 30 a not under their command.
In such a case, communications frequently occur between the CMs 30 a and 30 b. Thus, a higher load is imposed on the switch 302 a and CA301 a 2 of the CM30 a and the switch 302 b and CA 301 b 2 of the CM30 b.
As seen, when a copy process is performed, a higher load is imposed on particular locations in the system owing to another process. Further, a higher load is imposed on different locations depending on the situation of another process.
Where virtualization using logical volumes is not performed, the server 1 may previously store patterns having a higher load in each situation and assign a request to the virtualization switch device 2 a or virtualization switch device 2 b in accordance with the current processing situation.
On the other hand, where the disk units 305-1 to 305-5 are virtualized as in this embodiment, the virtualization switch devices 2 a and 2 b convert addresses of the logical volumes LV-01 and LV-02 into those of the physical disk units 305-1 to 305-8. For this reason, the server 1 may not know which of the disk units 305-1 to 305-8 is actually accessed in accordance with a request based on the specification of an address of the logical volume LV-01 or logical volume LV-02 and whether a communication is performed between the CMs 30 a and 30 b on the basis of the request.
For this reason, in this embodiment, when the server 1 receives a copy request, it conducts a copy test before performing an actual process to determine to which of the virtualization switch devices 2 a and 2 b the copy request may be sent. Hereafter, a process including such a copy test will be described with reference to FIG. 13.
When the processor 102 of the server 1 receives a copy request from the terminal 4 via the communication adaptor 104 (S1101 to S1102), it sends this copy request to the virtualization switch devices 2 a and 2 b. The processors 203 a and 203 b of the virtualization switch devices 2 a and 2 b receive this request and then refer to the logical volume management tables 205 a and 205 b. The processors 203 a and 203 b then determine that the virtualization switch device 2 a or virtualization switch device 2 b corresponding to the copy source logical volume in the logical volume management tables 205 a and 205 b is a master virtualization switch device. In this embodiment, it is assumed that the number of the copy source logical volume is 2 a and that the virtualization switch device 2 a has been determined to be the master.
Subsequently, the virtualization switch device 2 a, which has been determined to be the master, performs the process of S1104 (S1104). This process of S1104 will be described with reference to FIG. 14.
In this embodiment, the virtualization switch device-in-charge management tables 206 a and 206 b stored in the storage units 204 a and 204 b are tables for managing copy processes in progress and are illustrated in FIG. 15. Specifically, the virtualization switch device-in-charge management tables 206 a and 206 b assign a session ID to each copy process and are storing the virtualization switch device 2 a or virtualization switch device 2 b in charge of copy for each copy process. The virtualization switch device-in-charge management tables 206 a and 206 b are also storing the copy source virtual volume name and the copy destination virtual volume name for each copy process, as well as the average copy rate of each copy process in progress as copy performance (Mbps). The both path status of the virtualization switch device-in-charge management tables 206 a and 206 b shows whether a failure is present in one or both of the access paths to the disk device 3 via the virtualization switch devices 2 a and 2 b and shows that if the both path status is Online, communication is possible in both paths.
For example, a virtualization switch device-in-charge management table 105 of FIG. 15 illustrates, with regard to a copy process illustrated by a session ID “1,” that the virtualization switch device in charge of copy is 2 a and that the copy source virtual volume and the copy destination virtual volume are “LV01” and “LV-02,” respectively. The table also shows that the copy performance of the copy process illustrated by the session ID “1” is 500 Mbps. Note that when one of the virtualization switch device-in-charge management tables 206 a and 206 b is updated, the processors 203 a and 203 b communicate with each other via the ports 201 a 6 and 201 b 6 so that the other table has the same contents as the updated table. Hereafter, if not mentioned, it is assumed that when one of the virtualization switch device-in-charge management tables 206 a and 206 b is updated, an update process is always performed so that the other table has the same contents as the updated table.
In the flowchart of FIG. 14, the processor 203 a serving as the master refers to the virtualization switch device-in-charge-of-copy management table 206 a to check whether one of the virtualization switch devices 2 a and 2 b is in charge of no copy (has no copy processes in the table it is in charge of) (S2101). If one of the virtualization switch devices 2 a and 2 b is in charge of no copy, the processor 203 a instructs this virtualization switch device to take charge of copy as a virtualization switch device in charge of the current copy request (S2109). The processor 203 a then adds the selected virtualization switch device 2 b to the virtualization switch device-in-charge-of-copy management table 206 a and then returns from the process illustrated in FIG. 14 to the process illustrated in FIG. 13.
Subsequently, the virtualization switch device 2 b, which has been instructed to take charge of copy, performs an actual copy process (S1105).
This process allows transfer of the copy request to the low-load virtualization switch device (2 a or 2 b) without performing the copy test process illustrated in S2102 to S2108.
In contrast, as illustrated in FIG. 14, if both the virtualization switch devices 2 a and 2 b are performing any copy process when the processor 203 a refers to the virtualization switch device-in-charge-of-copy management table 105, the processor 203 a performs the processes of S2102 to S2108.
That is, first, the processor 203 a repeatedly issues a read command to the both the copy source address and the copy destination address contained in the copy request (S2102).
Subsequently, the virtualization switch device 2 a counts the respective numbers of responses within a predetermined time to the read commands issued to both the copy source address and the copy destination address and stores the smaller number as N1 (S2103). Although the location where the number is stored is not particularly specified, it is assumed that the number is stored in the storage area within the processor 102 or in a memory (not illustrated).
Since a copy process is a process of writing data in the copy source to the copy destination, the data may not be written unless reading of the data is completed, and next data may not be written unless writing of the previous data is completed. That is, even if one of the copy source and the copy destination is faster in processing speed than the other, the copy process may not be performed only at the slower processing speed of the other. Accordingly, the smaller counted number within the predetermined time, that is, the counted number of the slower is used.
Subsequently, through the port 201 a 6, the processor 203 a instructs the virtualization switch device 2 b to repeatedly issue a read command to both the copy source address and copy destination address contained the copy request (S2104).
The processor 203 b receives this instruction, counts the respective numbers of responses within a predetermined time, and sends the smaller number to the virtualization switch device 2 a as N2 via the port 201 b 6. The processor 203 a receives and stores N2 (S2105). Although the location where the number is stored is not particularly specified, it is assumed that the number is stored in the storage area within the processor 102 or in a memory (not illustrated). The reason why the smaller counted number is used is the same as that shown in the above-mentioned S2103.
The processor 203 a then compares the stored N1 and N2 (S2106). If the value of N1 is larger than N2, the processor 203 a determines that the virtualization switch device in charge of the copy request is 2 a, and continuously takes charge of the copy process. This is because the processor 203 a determines that the larger response number represents better transfer performance (S2107) In contrast, if the value of N2 is larger than N1, the processor 203 a instructs the virtualization switch device 2 b to perform the copy process.
After the processes of S2107 and S2108, the processor 203 a stores the virtualization switch device 2 a or virtualization switch device 2 b in charge of copy in the management table 206 a and then returns from the process illustrated in FIG. 14 to the process illustrated in FIG. 13.
Upon completion of the process of FIG. 14, the processor 203 a or processor 203 b of the virtualization switch device 2 a or virtualization switch device 2 b in charge of copy performs the copy process.
As described above, the server 1 may not know how the virtualization switch device 2 a or virtualization switch device 2 b is accessing the disk units 305-1 to 305-8, owing to the address conversion of the logical volume into the disk device performed by the virtualization switch device 2 a or virtualization switch device 2 b.
However, as in the above-mentioned process, the virtualization switch device serving as the master (2 a in this embodiment) conducts a test about the transfer speed to grasp the speed of the actual copy process and then selects 2 a or 2 b as a virtualization switch device to which the copy request is to be transferred. Thus, in the storage system according to this embodiment, it is possible to send the copy request to the virtualization switch device 2 a or virtualization switch device 2 b which can be accessed along a path having a lower load even when the above-mentioned address conversion is performed.
The case where the terminal 4 sends a new copy request has been described above. Monitoring the transfer speed of each copy process in progress in this process allows detection of a copy process exhibiting low copy performance and re-selection of the path of such a copy process.
This process will be described with reference to FIG. 16.
The processor 203 a of the virtualization switch device 2 a serving as the master is monitoring the transfer speed of each copy process in progress with reference to the virtualization switch device-in-charge management table 206 a (S3101).
If the processor 203 a receives no response from one of the virtualization switch devices (2 a or 2 b), it updates into “Offline” the both path status of the session ID the no-response virtualization switch device (2 a or 2 b) is in charge of.
The processor 203 a then checks whether all the paths are Online (S3102). If there is an Offline path, the path has a fault and thus an optimization process using a copy test may not be performed. Accordingly, the processor 203 a returns the process to S3101.
The processor 203 a then refers to the virtualization switch device-in-charge management table 206 a to check whether the copy process count (session count) has reached its maximum (S3103). In this embodiment, it is assumed that the maximum session count is “15.” If the session count has reached its maximum, the virtualization switch device in charge of any session may not be changed. Accordingly, optimization is not possible. For this reason, the processor 203 a does not perform subsequent processes and returns the process to S3101.
The processor 203 a also refers to the virtualization switch device-in-charge management table 206 a to check for unevenness of each parameter (S3104).
In this process, the processor 203 a checks whether there is a difference between the respective numbers of copy processes the virtualization switch devices 2 a and 2 b in charge of copy are performing and whether there is a copy process exhibiting lower performance than the other copy processes.
For example, when the virtualization switch device-in-charge management table 206 a shows statuses illustrated in FIG. 17, the number of copy processes the virtualization switch device 2 a is in charge of is 4. On the other hand, the number of copy processes the virtualization switch device 2 b is in charge of is 2. Thus, the processor 203 a confirms that copy processes are unevenly distributed to the virtualization switch device 2 a.
The processor 203 b also confirms that the transfer speed of a copy process having a session ID1 is low, as copy performance information.
The processor 203 a then determines whether the number of copy processes one of the virtualization switch devices 2 a and 2 b is in charge of is “0.”If there is a virtualization switch device of “0,” the processor 203 a determines whether there is a copy process in progress (S3106). If there is no copy process in progress, the virtualization switch device-in-charge management table 105 contains no item about a copy process in progress. Accordingly, the processor 203 a determines whether there is a copy process in progress by checking whether there is an item about a copy process in progress in the virtualization switch device-in-charge management table 206 a. If there is no copy process in progress, optimization is not done. Accordingly, the processor 203 a proceeds to the process of S3101. If the processor 203 a determines in the process of S3106 that there is a copy process in progress, it proceeds to the process of S3108.
Further, if the processor 203 a determines in S3105 that none of the virtualization switch devices 2 a and 2 b is “0,” it checks whether there is unevenness in copy performance (S3107). If there is no unevenness in performance, optimization is not done. Accordingly, the processor 203 a returns to the process of S3101. In this embodiment, the processor 203 a makes the determination about unevenness by determining whether there is a copy process exhibiting copy performance 30% lower than the average copy performance.
The above-mentioned unevenness determination method according to this embodiment is only illustrative and the processor 203 a may make the determination by determining whether there is a copy process exhibiting copy performance lower than the most common copy performance by a predetermined value or more. Alternatively, the processor 203 a may make the determination by determining whether that there is a copy process exhibiting copy performance lower than the highest copy performance by a predetermined value or more.
In contrast, if there is unevenness in performance, the processor 203 a performs the processes of S3108 and later steps.
Since the virtualization switch device-in-charge management table 105 b of FIG. 17 shows that the copy performance of a copy process having a session ID 1 is lower than the average copy performance (450 MBPS) by 30% or more, the processor 203 a determines that there is a copy process exhibiting copy performance lower than the average copy performance by 30% or more.
If the processor 203 a determines in S3106 or S3107 that it performs the processes of S3108 and later steps, it selects a copy process that another virtualization switch device, 2 a or 2 b, preferably takes charge of, from among the copy processes in progress (S3108).
Specifically, the processor 203 a makes the selection on the basis of a criterion that the numbers of copy processes the virtualization switch devices 2 a and 2 b are in charge of are approximately equal and a criterion that the change of the virtualization switch device in charge of copy, 2 a or 2 b, is selectively performed with respect to a copy process exhibiting low copy performance.
That is, in the virtualization switch device-in-charge management table 206 a of FIG. 17, the processor 203 a selects the copy process having the session ID 1 as a copy process that another virtualization switch device preferably take charge of.
The processor 203 a then suspends the copy process selected in S3108 (S3109). The processor 203 a then selects a virtualization switch device in charge of copy for the copy process (S3110). This process is the process illustrated in FIG. 14. That is, if an increase in copy performance can be expected by changing the virtualization switch device in charge of the copy process in this process, a process of changing the virtualization switch device in charge of copy (2 a or 2 b) is performed.
When this process is complete, the virtualization switch device in charge of copy selected in S3110 resumes the copy process suspended in S3109 (S3111). The processor 203 a proceeds to the process of S3101.
In selecting the virtualization switch device in charge of copy, 2 a or 2 b, in S3108, multiple copy processes may exhibit copy performance falling below a predetermined value, as shown in a virtualization switch device-in-charge management table 105 c of FIG. 18. In this case, the processor 203 a may realize optimization by collectively handling the multiple copy processes and performing the process of S3110 on each copy process.
All examples and conditional language recited herein are intended for pedagogical purposes to aid the reader in understanding the invention and the concepts contributed by the inventor to furthering the art, and are to be construed as being without limitation to such specifically recited examples and conditions, nor does the organization of such examples in the specification relate to a showing of the superiority and inferiority of the invention. Although the embodiments of the present invention have been described in detail, it should be understood that the various changes, substitutions, and alterations could be made hereto without departing from the spirit and scope of the invention.

Claims

1. A virtualization control apparatus of a storage system, the storage system including a plurality of storage controllers controlling a plurality of storage units connected thereto and a plurality of virtualization switch units being connected to one of the storage controllers, each of the virtualization switch units converting a copy source logical volume address and a copy destination logical volume address contained in a received copy request into addresses of the storage units and making a request to the storage controllers on the basis of the converted addresses, the virtualization control apparatus comprising:

a selection unit that, when receiving a copy request, conducting an access test corresponding to the copy request on each of the virtualization switch units and selects one of the virtualization switch units of the highest performance among the plurality of the virtual switch units; and

a sending unit that sends the copy request to the selected virtualization switch unit.

2. The virtualization controller according to claim 1, wherein

the selection unit sends requests for reading a copy source address and a copy destination address contained in a received copy request to the virtualization switch units, then conducts an access test, the access test being a test where the lower of response performance to the copy source read request and response performance to the copy destination read request sent by each virtualization switch unit is compared with each other, and selects a virtualization switch unit exhibiting the highest response performance among the virtualization switch units in the access test.

3. The virtualization controller according to claim 1, wherein,

when any of the virtualization switch units is performing no copy process when the selection unit receives the copy request, the selection unit selects the virtualization switch unit performing no copy process, without conducting the access test.

4. A virtualization control apparatus of a storage system, the storage system including a plurality of storage controllers that controls a plurality of storage units connected thereto and a plurality of virtualization switch units being connected to one of the storage controllers, each of the virtualization switch units converting a copy source logical volume address and a copy destination logical volume address contained in a received copy request into addresses of the storage units and making a request to the storage controllers on the basis of the converted addresses, the virtualization control apparatus comprising:

a monitoring unit that monitors unevenness in copy performance of copy processes which process by each of the virtualization switches;

a selection unit when detecting unevenness in the copy performance, suspending a copy process exhibiting copy performance lower than predetermined copy performance among the copy processes the virtualization switch units are performing, conducting an access test corresponding to the suspended copy process on each of the virtualization switch units, and selecting one of the virtualization switch units of the highest performance among the plurality of the virtual switch units; and

a processing unit that resumes the suspended copy process using the selected virtualization switch unit.

5. The virtualization controller according to claim 4, wherein

the selection unit sends requests for reading a copy source address and a copy destination address of the suspended copy process to the virtualization switch units, then conducts an access test, the access test being a test where the lower of response performance to the copy source read request and response performance to the copy destination read request sent by each virtualization switch unit is compared with each other, and selects a virtualization switch unit exhibiting the highest response performance among the virtualization switch units in the access test.

6. The virtualization controller according to claim 4, wherein,

if any of the virtualization switch units is performing no copy process when the selection unit suspends the copy request, the selection unit selects the virtualization switch unit performing no copy process, without conducting the access test.

7. A non-transitory computer-readable medium storing a program applied for a virtualization control apparatus of a storage system, the storage system including a plurality of storage controllers controlling a plurality of storage units connected thereto and a plurality of virtualization switch units being connected to one of the storage controllers, each of the virtualization switch units converting a copy source logical volume address and a copy destination logical volume address contained in a received copy request into addresses of the storage units and making a request to the storage controllers on the basis of the converted addresses, the program causing the virtualization control apparatus to execute:

conducting, when receiving a copy request, an access test corresponding to the copy request on each of the virtualization switch units, and selecting one of the virtualization switch units of the highest performance among the plurality of the virtual switch units; and

sending the copy request to the selected virtualization switch unit.